We had a really interesting discussion yesterday about voting on Lemmy/PieFed/Mbin and whether they should be private or not, whether they are already public and to what degree, if another way was possible. There was a widely held belief that votes should be private yet it was repeatedly pointed out that a quick visit to an Mbin instance was enough to see all the upvotes and that Lemmy admins already have a quick and easy UI for upvotes and downvotes (with predictable results ). Some thought that using ActivityPub automatically means any privacy is impossible (spoiler: it doesn’t).
As a response, I’m trying this out: PieFed accounts now have two profiles within them - one used for posting content and another (with no name, profile photo or bio, etc) for voting. PieFed federates content using the main profile most of the time but when sending votes to Mbin and Lemmy it uses the anonymous profile. The anonymous profile cannot be associated with its controlling account by anyone other than your PieFed instance admin(s). There is one and only one anonymous profile per account so it will still be possible to analyze voting patterns for abuse or manipulation.
ActivityPub geeks: the anonymous profile is a separate Actor with a different url. The Activity for the vote has its “actor” field set to the anonymous Actor url instead of the main Actor. PieFed provides all the usual url endpoints, WebFinger, etc for both actors but only provides user-provided PII for the main one.
That’s all it is. Pretty simple, really.
To enable the anonymous profile, go to https://piefed.social/user/settings and tick the ‘Vote privately’ checkbox. If you make a new account now it will have this ticked already.
This will be a bit controversial, for some. I’ll be listening to your feedback and here to answer any questions. Remember this is just an experiment which could be removed if it turns out to make things worse rather than better. I’ve done my best to think through the implications and side-effects but there could be things I missed. Let’s see how it goes.
You must log in or register to comment.
Cool solution. It’s great to have multiple projects in the fediverse that can experiment with different features/formats.
For those who are concerned about possible downsides, I think it’s important to understand that
- PieFed has a small userbase
- Rimu is an active admin, so if you are attempting to combat brigading or other bad behavior and this makes it more difficult, just send them a DM and they will be happy to help out
This is a good environment to test this feature because Rimu can keep a close watch over everything. We can’t become paralyzed by the hypothetical ways that bad actors might abuse new features or systems. The only way forward is through trial and error, and the fact that PieFed exists makes that process significantly faster and less disruptive.
This is an attempt to add more privacy to the fediverse. If the consequences turn out for the worse, then we can either try something else, or live with the lack of privacy. Either way, we’ll be better off than having never tried anything at all.
Just upvoted myself but nobody else knows 🤫
Edit: Actually I forgot to toggle the setting before voting on my own comment, so admins will see my @imaqtpie@piefed.social account upvoted the parent comment. Worth noting that it needs to be manually enabled.
Then I turned the setting on and voted on a bunch of other comments in this post. My anonymized voting account appears as @hED5TzoZomb@piefed.social, admins should be able to see it by checking the votes in this thread.
Point being, you can still track serial downvoters and harassment just as easily. But now you will need to take an extra step and message the instance admin (Rimu) and ask that they either reveal the identity of the linked profile or deal with it themselves. And that’s a good thing, imho.
You don’t even need to message an admin. You can just ban the agent doing the voting.
Point being, you can still track serial downvoters and harassment just as easily. But now you will need to take an extra step and message the instance admin (Rimu) and ask that they either reveal the identity of the linked profile or deal with it themselves. And that’s a good thing, imho.
This puts the privacy shield in the hands of a users instance admin. I like that approach, but I’m sure others will disagree.
This is more or less how it worked on Reddit. The admins handled vote spam or abuse, there was absolutely no expectation for moderators to have that information because the admins were dealing with the abuse cases. Moderators only concerned themselves with content and comments, the voting was the heart of how the whole thing works, and therefore only admins could see and affect them. Least privilege, basically.
I think a side effect of this, though, is that it increases the responsibility on admins to only federate with instances that have active and cooperative admins. It increases their responsibilities and demands active monitoring, which isn’t a bad thing, but I worry about how the instances that federates openly by default will continue to operate.
If you have to trust the admins, how do you handle new admins, or increasingly absent ones? What if their standards for what constitutes “harassment” don’t match yours? Does the whole instances get defederated? What if it’s a large instance, where communities will be cut off?
I don’t ask any of this as a way to put down this effort because I very, very much want to see this change, but there’s gonna be hurtles that have to be overcome
Ultimately I think the best solution would need assistance from the devs but I’m lieu of that, we have to make due.
You keep delivering, thank you so much!
That’s pseudonymous!
But all kidding aside, it sounds good
Awesome! This is the exact stopgap implementation I was arguing for, and I’m surprised how many people kept insisting it was impossible. You should try and get this integrated into mainline Lemmy asap. Definitely joining piefed in the meantime though.
While not a perfect solution, this seems very smart. It’s a great mitigation tactic to try to keep user’s privacy intact.
Seems to me there’s still routes to deanonymization:
- Pull posts that a user has posted or commented in
- Do an analysis of all actors in these posts. The poster’s voting actor will be over represented (if they act like I assume most users do. I upvote people I reply to etc)
- if the results aren’t immediately obvious, statistical analysis might reveal your target.
Piefed is smaller than lemmy, right? So if only one targeted posting account is voting somewhat consistently in posts where few piefed users vote/post/view, you got your guy.
Obviously this is way harder than just viewing votes. Not sure who would go to the trouble. But a deanonymization attack is still possible. Perhaps rotate the ids of the voting accounts periodically?
It will never be foolproof for users coming from smaller instances, even with changing IDs. If you see a downvote coming from PieFed.social you already have it narrowed down to not too many users, and the rest you can probably infer based on who contributes to a given discussion.
Still, I think it’s enough to be effective most of the time.
You’re a hero for making this happen in… 24 hours? 48?
The issue won’t go away, we’ll see how well everyone else deals with it, but this is a super strong argument for your system / server.
(Advertise it. Advertise it HARD. “piefed, we have private votes”.)
It would be useful for a community to have the ability opt-out of Private voting. i.e. some meta information in the sidebar that piefed can see and it will not enable private voting for that community.
That’s super cool and amazing that you implemented it so quickly.
So now I have a PieFed account :)
This is quite a smart solution, good job !
Hey, Lemmy admin here. If I ban an anonymous account, does the account it’s tethered to also get banned?
No but perhaps it should!
PieFed lacks an API, making it an unattractive tool for scripting bots with. I don’t think you’ll see any PieFed-based attacks anytime soon.
What about PieFed-based shitty humans?
PieFed tracks the percentage of downvotes vs upvotes (calling it “Attitude” in the code and admin UI ), making it easy to spot people who downvote excessively and easy to write functionality that deals with them. Perhaps anonymous voting should only be available to accounts with a normal attitude (within a reasonable tolerance).
Wow your documentation is so much better than ours.
Do you have a link? The Piefed docs page is empty for me.
Ah fuck! I mistook the piefed docs for the pixelfed docs.
Yes but … Navigation icon at the top right of the pages leads to these :
https://join.piefed.social/docs/piefed-mobile/ https://join.piefed.social/docs/developers/ https://join.piefed.social/docs/admin-guide/ https://join.piefed.social/docs/installation/
https://join.piefed.social/2024/06/22/piefed-features-for-growing-healthy-communities/I swear there was documentation there.
That’s nice of you to say. I’ve tried to focus well on certain areas that seem important but I really admire the breadth of https://join-lemmy.org/docs/ which I could never hope to cover.
Do you ben based on voting behaviour?
If the same account is voting in the same direction on every single post and comment in an entire community in a matter of seconds while contributing neither posts nor comments? Yes, vote manipulation.
If one user is following another around, down voting their content across a wide range of topics? Yes, targeted harassment.
Sure, but by the same token, mods are just as capable of manipulation and targeted harassment when they can curate the voting and react based on votes.
On reddit, votes are only visible to the admins, and the admins would take care of this type of thing when they saw it (or it tripped some kind of automated something or other). But they still had the foresight not to let moderators or users see those votes.
Complete anonymity across the board won’t work but they’re definitely needs to be something better than it is now.
mods are just as capable of manipulation and targeted harassment when they can curate the voting and react based on votes
I’m not sure what you’re trying to say.
I’m speaking as an admin, not as a mod. I own the servers. I have direct access to the databases. When law enforcement comes a’knockin’, it’s my ass that gets arrested. I have total control over my instances and can completely sever them from the fediverse if I feel it necessary. Mods are mall cops that can lock posts and deal with problem users one at a time.
On reddit, votes are only visible to the admins, and the admins would take care of this type of thing when they saw it (or it tripped some kind of automated something or other)
There are no built in automations. Decoupling votes from the users that cast them interferes with my ability to “take care of this type of thing.”
Yeah, I see that and it does concern me now that it has been brought up.
However. In the last 6 months of being active in the ‘Lemmy.world defense hq’ matrix room where we coordinate admin actions against bad people, vote manipulation has come up once or twice. The other 99% of the time it’s posts that are spam, racist or transphobic. The vote manipulation we found detected using some scripts and spreadsheets, not looking at the admin UI. After all, using code is the only way to scan through millions of records.
Downvote abuse/harassment coming from PieFed will be countered by monitoring “attitude” and I have robust tools for that. I can tell you with complete confidence that not one PieFed user downvotes more than they upvote. I can provide 12 other accounts on Lemmy instances that do, tho. Lemmy’s lack of a similar admin tool is unfortunate but not something I can do anything about.
What I’ve done with developing this feature is taken advantage of a weakness of ActivityPub - anyone can make accounts and have them do stuff. Even though I’ve done it in a very controlled and limited way and released all the code for it, having this exposed feels pretty uncomfortable. There were many many people droning on about “votes must be public because they need to come from an account” blah blah and that secure safe illusion has been ripped away now. That sucks, but we were going to have to grapple with it eventually one way or another.
Anyway. I’m not wedded to this or motivated by a fixed ideology (e.g. privacy über alles) so removing this is an option. It didn’t even take that long to code, I spent more time explaining it than coding it.
Would banning the voting half of the pseudonymous account not mitigate the immediate issue? Then asking their instance admin to later lookup and ban the associated commentating account.
Well, doesn’t that fly in the face of federated autonomy and privacy?
On one end, if it’s my instance and I want to ban a user, I want the whole fucking user banned – not just remove their ability to vote anonymously. If one of my communities or users is being attacked, it’s my responsibility to react. If I can’t remove the whole problem with a ban, then I have to remove the whole problem with a de-federation. (A thing I fundamentally don’t want to do.)
On the other, if some other admin says, “one of your users is being problematic, please tell me who they are,” I’m going to tell that other admin to fuck right off because I just implemented a feature that made their votes anonymous. I’m not about to out my users to some rando because they’re raining downvotes on MeinHitler69@nazi.hut.
It’s a philosophical difference of opinion.
On one end, if it’s my instance and I want to ban a user, I want the whole fucking user banned – not just remove their ability to vote anonymously.
I mean, is that truly the case? If a user only engages in vote manipulation, but otherwise they have insightful comments/posts, is it really that big of a deal that you will ban only their option to vote?
But if the only bad behavior is voting and you can that agent then you’ve solved the core issue. The utility is to remove the bad behavior, no?
No, the utility is to remove bad users.
To prevent them from engaging in bad behavior.
I use people upvoting bigoted and transphobic content to help locate other bigoted and transphobic accounts so I can instance ban them before they post hate in to our communities.
This takes away a tool that can help protect vulnerable communities, whilst doing nothing to protect them.
It’s a step backwards
Yea, which is why I think the obvious solution to the whole vote visibility question is to have private votes that are visible to admins and mods for moderation purposes. It seems like the right balance.
It will be difficult to get the devs of Lemmy, Mbin, Sublinks, FutureProject, SomeOtherProject, etc to all agree to show and hide according to similar criteria. Different projects will make different decisions based on their values and priorities.
…and it still doesn’t solve the issue that literally anyone can run their own instance and just capture the data.
The OP discusses exactly a solution to the anyone setting up an instance to capture the data, because the users home instance federates their votes anonymously.
There maybe flaws in it, not that’s exactly what it aims to solve.
whilst doing nothing to protect them
Well it also takes away a tool that harassers can use for their harassing of individuals, right? This does highlight the often-requested issue of Lemmy needs better/more moderation tools though.
It actually adds a tool for harassers, in that targeted harassment can’t be tied back to a harasser without the cooperation of their instance admin.
In reality, I think a better answer might be to anonymize the username and publicize the votes.
Hmm, yes.
PieFed tracks the percentage of downvotes vs upvotes (calling it “Attitude” in the code and admin UI), making it easy to spot people like this and easy to write functionality that deals with them. Perhaps anonymous voting should only be available to accounts with a normal attitude (within a reasonable tolerance).
PieFed tracks the percentage of downvotes vs upvotes (calling it “Attitude” in the code and admin UI)
That’s cool. I wonder what my attitude is and I wonder how accurate the score is, if our federations don’t overlap super well. What happens if I have a ton of interactions on an instance that yours is completely unaware of?
(I think “Attitude” is a perfect word, because it’s perceptive. Like, “you say they’re great but all I see them do is get drunk and complain about how every Pokemon after Mewtwo isn’t ‘legit’,” sort of thing.)
I’ve intentionally subscribed to every active community I can find (so I can populate a comprehensive topics hierarchy ) making piefed.social get a fairly complete picture. Your attitude is only 3% below the global average, nowhere near the point where I’d take notice.
If public voting data becomes a thing across the threadiverse, as some lemmy people want.
Which is why I think the appropriate balance is private votes visible to admins/mods.
Admins only. Letting mods see it just invites them to share it on a discord channel or some shit. The point is the number of people that can actually see the votes needs to be very small and trusted, and preferably tied to a internal standard for when those things need acted upon.
The inherent issue is public votes allow countless methods of interpreting that information, which can be acted on with impunity by bad actors of all kinds, from outside and within. Either by harassment or undue bans. It’s especially bad for the instances that fuck with vote counts. Both are problems.
I’m going to have to come up with set criteria for when to de-anonomize, aren’t I. Dammit.
In the meantime, get in touch if you spot any bigot upvotes coming from PieFed.social and we’ll sort something out.
The problem is, it’s more than just the upvote. I don’t ban people for a single upvote, even on something bigoted, because it could be a misclick. What I normally do is have a look at the profiles of people who upvote dogwhistle transphobia, stuff that many cis admins wouldn’t always recognise. And those upvotes point me at people’s profiles, and if their profile is full of dog whistles, then they get pre-emptively instance banned.
So you can still ban the voting agent. Worst case scenario you have to wait for a single rule breaking comment to ban the user. That seems like a small price to pay for a massive privacy enhancement.
Ahh, right, got it.
Let’s keep an eye on this. I am hopeful that with PieFed being unusually strong on moderation in other respects that we don’t harbor many people like that for long.
This is great
@rimu is there a forum style ap implementation that can talk to lemmy communities (I’m assuming that piefed can) without voting?
I believe NodeBB is what you’re looking for . https://community.nodebb.org/
@Dame Ty!
Oh god…I’m Charlie Kelly.
I read that as “Pirate voting”.
How does this work with moderation? I.e. what happens if I ban the real user from a Lemmy instance? What if I ban the alternate user?
Also, what happens if on Piefed, a user votes for something, then they change the setting and then they vote for the same thing again? How would a Lemmy instance know if it should count the vote or not, since the original user didn’t actually vote from Lemmy’s point of view?
The ‘real user’ and the ‘private voter’ are 2 different accounts as far a external instances are concerned, but only 1 as far as piefed.social is concerned. So if you banned either one, it would have the same effect, because PF would locate the same account from the information provided.
Likewise, a piefed user can’t vote twice on something, they make one vote, and then the ‘private voting’ setting determines how it is sent out. The local system has tracked that they have voted, and changing the setting won’t change that.
There’s always more work to do of course, but piefed.social is a small instance, with manual approval required for registration, no API to script things like mass downvoting, and concepts such as ‘attitude’ which would prevent that anyway, so I can’t foresee anything too disastrous happening from this little experiment.
I’m a little concerned about the precedent this sets. An instance could use this technique to facilitate anonymous commenting or posting in addition to votes.
Who cares? Generating an infinite number of tokenized identities to facilitate ban evasion will just result in an instance getting defederated. This introduces no real risk as long as the instance is generally abiding by the rules.
Most of us here are fairly anonymous anyway. I dont think being able to add an additional layer of privacy to our activity is really a big deal.
I missed the discussion on voting the other day it seems, but for what it’s worth, I like the voting system. In real life discussions happen in open air, and don’t hang there in posterity for people to stumble upon after. When we come to a consensus in conversation it is then left at that and we move on.
When online, these discussions stay as they are, and I think voting gives a way of people to come to a consensus, to leave a mark upon the conversation such that the people who come behind understand how everyone felt about it.
This is helpful I think, because it does not hide the down votes on nasty comments or ideas that hurt others.
One of the most interesting and horrible things about the internet is that every village has a “crazy Bob” but because they were the minority the good of the people outnumbered their outlandish or hateful ideas.
Now they can and do find each other online, forming a vocal and damaging minority. Without the majority able to show their dislike, human nature means more will fall in line with them and their ideals.
